Agent sse communication method based on multi-party threshold derivation and business causal dag

Abstract

The present application relates to the technical field of logistics communication security, and in particular to an agent SSE communication method and system based on multi-party threshold derivation and business causal DAG, wherein a scene sensitivity parameter based on a session is used to dynamically determine a threshold threshold t of a Shamir threshold secret sharing scheme, a business causal hash DAG is constructed, and a zero-knowledge scene attribute proof ZK-SAP is used to bind the shard legality to the reachable state transition path of the business state machine, the zero-knowledge scene attribute proof ZK-SAP is embedded into an extension header, so that the server can complete the access verification without obtaining the plaintext of the user sensitive attribute, thereby avoiding the privacy exposure risk caused by the attribute plaintext transmission or signature verification in the existing scheme, and finally, the shard length distribution tends to be uniformly distributed through Rényi entropy adaptive padding, so that an attacker cannot speculate the business type by analyzing the shard length mode.

Description

Technical Field

[0001] This invention relates to the field of logistics communication security technology, and more specifically, to an agent SSE communication method based on multi-party threshold derivation and business causal DAG. Background Technology

[0002] Current intelligent logistics agent conversational systems, such as intelligent order placement, digital scheduling, intelligent billing, and supply chain assistants, typically use streaming interfaces to enable real-time interaction between users and intelligent agents. Their core shortcomings lie in security and scenario adaptability, as detailed below: 1. The native SSE protocol lacks a business-level encryption mechanism: Sensitive information such as cargo value, loading and unloading address, vehicle trajectory, driver's license, contract billing rules and settlement account entered by the cargo owner may be transmitted in plaintext through the SSE's data field or streaming response. Attackers can obtain this information through packet sniffing or man-in-the-middle methods, leading to the leakage of logistics business privacy. 2. Poor adaptability of key management to logistics flow scenarios: Existing solutions mostly use fixed API tokens, long-term application keys or single-point server keys; once the token or long-term key is leaked, attackers can impersonate cargo owners, carriers or dispatchers to initiate sessions such as smart order placement, waybill inquiry, and fee adjustment. Moreover, key replacement requires the synchronization of multiple front-ends, mini-programs, gateways and business services, which can easily cause service interruptions. 3. Weak anti-replay and anti-forgery capabilities of streaming fragments: SSE fragment transmission is asynchronous and lacks causal binding with the logistics business status. Attackers can intercept historical fragments and replay them at other stages, such as replaying fragments from the quotation stage to the settlement stage, or forging fragments such as vehicle location, receipt, return receipt, and fee confirmation to insert into the stream, causing the agent to perform incorrect business actions. 4. Disconnect between security strategy and sensitivity level of logistics conversation: If the same security strategy is used in different scenarios, such as pre-sales consultation, ordinary order placement, transportation of hazardous chemicals, transportation of high-value goods, cross-park warehousing, and reconciliation settlement, unnecessary delays will be added in low-sensitivity scenarios, while high-sensitivity scenarios may be insufficiently protected, lacking the ability to dynamically adapt to the value of goods, hazard level, route risk and number of participating entities. Summary of the Invention

[0003] The purpose of this invention is to provide an agent SSE communication method based on multi-party threshold derivation and business causal DAG to solve the problems mentioned in the background art.

[0004] To address the aforementioned technical problems, one objective of this invention is to provide an agent SSE communication method based on multi-party threshold derivation and business causal DAG, comprising: S1: Obtain the scenario sensitivity parameters of the session, and dynamically determine the threshold t of the Shamir threshold secret sharing scheme based on the scenario sensitivity parameters; S2: Generate a session key DK. Based on the threshold t and the multi-party threshold key derivation MTKD mechanism, the session key DK is split into n key shares, and the n key shares are distributed to different cooperating entities. S3: Construct a business causal hash DAG; S4: Generate zero-knowledge scenario attribute proofs ZK-SAP; S5: Based on the Rényi entropy adaptive fragmentation and padding EAFL mechanism, the padding length is dynamically determined and the padding operation is performed according to the length of the current plaintext data to be sent and the preset padding length probability distribution target. S6: Using the session key DK, perform symmetric encryption on the padded plaintext data to generate an encrypted payload; S7: Assemble improved SSE frames; S8: Send the SSE frame via the SSE protocol.

[0005] Preferably, the threshold value t is positively correlated with the scene sensitivity parameter, and the formula for calculating the threshold value t is: ; Where n is the total number of cooperating entities participating in the key share distribution, σ(·) is the Sigmoid function used to map the comprehensive risk score to the interval between 0 and 1, w1~w4 are weight coefficients preset or trained based on historical waybills, vehicle trajectories, warehousing and settlement data; CargoValue is the normalized index of the value of goods corresponding to the current logistics session, HazardLevel is the cargo hazard level index, RouteRisk is the transportation route risk index, and PartyCount is the number of entities participating in this business collaboration.

[0006] Preferably, the derivation formula of the key DK in step S2 is: ; Where, λ i(0) represents the Lagrange interpolation coefficients, R represents the set of real-time online partners that have passed identity verification, t represents the threshold, and DK represents the session key; HKDF represents the key derivation function based on hash message authentication code, SHA3 represents the SHA3-512 secure hash algorithm, SortAttr(A) represents the information string obtained by sorting and normalizing the attribute list A according to the preset field encoding, and Nonce‖TimeSlot represents the salt value formed by concatenating the random number Nonce with the current time slice TimeSlot; any share leak less than t cannot restore DK, and the attribute list A only participates in the derivation as the Info parameter of HKDF but is not transmitted in plaintext.

[0007] Preferably, the nodes of the DAG represent fragments of the SSE protocol, and the directed edges of the DAG represent the legal state transition relationships between adjacent fragments based on a preset service state machine. The construction rule of the DAG is as follows: for the k-th partition Status The set of parent nodes is:

[0008] In the fragmentation extension header ; Wherein, Parents(F_k) is the set of parent nodes of the k-th SSE fragment F_k in the business causal hash DAG, used to represent the verified fragments that can serve as the basis for F_k in terms of time sequence and business state transition; e is the business event identifier parsed from the logistics business content carried by the fragment; H(C_j) is the digest value obtained by performing a secure hash operation on the complete frame content C_j of the parent fragment F_j, where C_j includes a fixed header, an extended header, an encrypted payload, and an authentication tag.

[0009] The verification rule for the DAG is as follows: when the current state of the server's maintenance state machine is obtained... At that time, received At that time, it was required All parent fragments have been verified and For a certain If valid, otherwise discard.

[0010] Preferably, the ZK-SAP proof is used to prove that the service attributes of the current session meet a preset admission policy without disclosing the attribute plaintext, and the constraint predicate of the ZK-SAP proof is... Defined as: ; Among them, CargoValue is the value of the goods, HazardLevel is the hazard level, RouteRisk is the route risk, PartyCount is the number of participants, VehicleType is the vehicle type, License is the vehicle or driver's license information, Route is the transportation route information, RestrictedZone is the restricted zone information, and DriverLicenseExpiry is the driver's license expiry date. An empty set indicates that the transport route does not intersect with the restricted area; ValidLicenses is a set of valid licenses and permits on the platform, including vehicle road transport permits, driver qualification certificates, carrier operating licenses, and hazardous or special transport qualifications, which are written into the Merkle tree after being reviewed by the platform, filed with the road administration, or verified by a third-party interface.

[0011] Preferably, the objective function of the Rényi entropy is: ; Where L is a random variable representing the slice length. It is second order entropy; The formula for calculating the fill length is as follows: ; in, It is the fill length. It is the plaintext length. It is an empirical distribution. It is a set of length buckets, and B = {256, 512, 1024, 2048, 4096}; The padding bytes of the Rényi entropy are generated from a DK-derived PRF, with a padding length l_pad=L. m, and write l_pad to the EAFL Control field of the extension header. After decryption, extract the valid bytes according to the plaintext length m.

[0012] Preferably, step S4 includes the following steps: S4.1: Compile the preset business access policy into an arithmetic circuit to generate the corresponding proof key and verification key; S4.2: The client obtains its own business attribute witness and uses the proof key and Groth16 zero-knowledge proof algorithm to generate a fixed-size proof π; S4.3: Write the proof π into the extended header of the SSE frame.

[0013] Preferably, the SSE frame includes a fixed header, an extended header, and an encrypted payload; The fixed header carries the encrypted packets of the n key shares, and the extended header carries the pointer set of the business causal hash DAG, the zero-knowledge scenario attribute proof, and the Rényi entropy filling control information. The SSE frame structure is as follows: .

[0014] On the other hand, this application also provides an agent SSE communication system based on multi-threshold derivation and business causal DAG, which, when applied to any of the foregoing methods, includes: The client and server are connected for communication. The client includes: The Multi-Party Threshold Dynamic Key Derivation (MTKD) module is used to dynamically determine the threshold threshold t for Shamir threshold secret sharing based on the scenario sensitivity parameter, and to split the session key DK into n key shares and distribute them to different partner entities. Business Causal Hash DAG Module; used to establish hash dependency topology between SSE shards based on a preset business state machine; The ZK-SAP module for zero-knowledge scenario attribute proofs is used to generate zero-knowledge proofs that demonstrate that business attributes meet admission policies. The Rényi entropy adaptive piecewise filling (EAFL) module is used to dynamically determine the filling length based on the second-order Rényi entropy target. A frame assembly and encryption module; used to construct an improved SSE frame containing a fixed header, an extended header, and an encrypted payload, and to configure the fixed header to carry the n key shares.

[0015] Preferably, the server includes: A frame parsing and decryption module is used to parse the fixed header, reconstruct the session key DK, and decrypt the encrypted payload. Causality verification module; used to verify the service continuity of the fragment sequence based on the DAG pointer set and the service state machine; Zero-knowledge verification module; used to verify the zero-knowledge proof to confirm the compliance of business attributes.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By coupling the threshold t with the scene sensitivity, lightweight protection for low-sensitivity scenes and strong security protection for high-sensitivity scenes are achieved. The session key DK is not completely stored on any single node. Even if an attacker breaks the client or a partner, they can only obtain one share and cannot reconstruct the DK. Moreover, the minimum number of shares required for reconstruction dynamically increases with the sensitivity, which significantly reduces the risk spread range after key leakage. 2. By using a business causal hash DAG, the sharding validity is bound to the reachable state transition path of the business state machine. The extension header is embedded through zero-knowledge scenario attribute proof ZK-SAP, which enables the server to complete access verification without obtaining the plaintext of user sensitive attributes, thus avoiding the privacy exposure risks caused by plaintext transmission of attributes or signature verification in existing solutions. 3. Rényi entropy adaptive padding makes the fragment length distribution nearly uniform, making it impossible for attackers to infer the business type by analyzing the fragment length pattern. This reduces the business identification rate from over 85% to below 12%, effectively combating traffic analysis attacks. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an agent SSE communication method based on multi-party threshold derivation and business causal DAG. Figure 2 This is a schematic diagram of the framework of an intelligent agent SSE communication system based on multi-party threshold derivation and business causal DAG. Attached image labels: 100, client; 200, server. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 like Figure 1 As shown, one of the objectives of this invention is to provide an agent SSE communication method based on multi-party threshold derivation and business causal DAG, comprising: S1: Obtain the scenario sensitivity parameters of the session, and dynamically determine the threshold t of the Shamir threshold secret sharing scheme based on the scenario sensitivity parameters; S2: Generate a session key DK. Based on the threshold t and the multi-party threshold key derivation MTKD mechanism, the session key DK is split into n key shares, and the n key shares are distributed to different cooperating entities. S3: Construct a business causal hash DAG; S4: Generate zero-knowledge scenario attribute proofs ZK-SAP; S5: Based on the Rényi entropy adaptive fragmentation and padding EAFL mechanism, the padding length is dynamically determined and the padding operation is performed according to the length of the current plaintext data to be sent and the preset padding length probability distribution target. S6: Using the session key DK, perform symmetric encryption on the padded plaintext data to generate an encrypted payload; S7: Assemble improved SSE frames; S8: Send the SSE frame to the agent server via the SSE protocol.

[0020] It should be noted that this application dynamically determines the threshold threshold t of the Shamir threshold secret sharing scheme based on the scenario sensitivity parameter of the session. By constructing a business causal hash DAG and zero-knowledge scenario attribute proof ZK-SAP, the legality of sharding is bound to the reachable state transition path of the business state machine. The zero-knowledge scenario attribute proof ZK-SAP is embedded in the extension header, which enables the server to complete the access verification without obtaining the plaintext of the user's sensitive attributes. This avoids the privacy exposure risk caused by the plaintext transmission of attributes or signature verification in existing schemes. Finally, the Rényi entropy adaptive filling makes the shard length distribution approach a uniform distribution, so that attackers cannot infer the business type by analyzing the shard length pattern.

[0021] In one embodiment of this application, the threshold t is positively correlated with the scene sensitivity parameter, and the calculation formula for the threshold t is: ; Where n is the total number of cooperating entities participating in the key share distribution, σ(·) is the Sigmoid function used to map the comprehensive risk score to the interval between 0 and 1, w1~w4 are weight coefficients preset or trained based on historical waybills, vehicle trajectories, warehousing and settlement data; CargoValue is the normalized index of the value of goods corresponding to the current logistics session, HazardLevel is the cargo hazard level index, RouteRisk is the transportation route risk index, and PartyCount is the number of entities participating in this business collaboration.

[0022] It should be noted that the preset weight coefficients are trained based on waybill data, making the sensitivity assessment more in line with the actual business risk distribution. Since existing technologies only switch sensitivity binary at the encryption algorithm level, this application maps sensitivity to the Shamir threshold parameter t, a continuously adjustable mathematical structure. By coupling the threshold t with the scenario sensitivity, lightweight protection for low-sensitivity scenarios and strong security protection for high-sensitivity scenarios are achieved, thereby selecting the appropriate threshold t according to different scenarios.

[0023] The client obtains a set R of online parties by querying the online status of each partner entity in real time (such as heartbeat detection or polling), where |R|≥t, and then requests the share held by each entity in set R. Calculate the Lagrange coefficients after receiving the data. Then, the reconstructed key is calculated, and finally, the IKM, salt value, and information string are input into HKDF-SHA3-512 to derive the final session key DK.

[0024] In one embodiment of this application, the derivation formula of the key DK in step S2 is: ; Where, λ i (0) represents the Lagrange interpolation coefficients, R represents the set of real-time online partners that have passed identity verification, t represents the threshold, and DK represents the session key; HKDF represents the key derivation function based on hash message authentication code, SHA3 represents the SHA3-512 secure hash algorithm, SortAttr(A) represents the information string obtained by sorting and normalizing the attribute list A according to the preset field encoding, and Nonce‖TimeSlot represents the salt value formed by concatenating the random number Nonce with the current time slice TimeSlot; any share leak less than t cannot restore DK, and the attribute list A only participates in the derivation as the Info parameter of HKDF but is not transmitted in plaintext.

[0025] It should be noted that in the key DK derivation formula... The scenarios for different roles are as shown in Table 1, the Role-Share Binding Table:

[0026] Shamir polynomial exist . share.

[0027] In one embodiment of this application, the nodes of the DAG represent fragments of the SSE protocol, and the directed edges of the DAG represent the legal state transition relationships between adjacent fragments based on a preset service state machine. The construction rule of the DAG is as follows: for the k-th partition Fk, the state is... The set of parent nodes is:

[0028] In the fragmentation extension header ; Wherein, Parents(F_k) is the set of parent nodes of the k-th SSE fragment F_k in the business causal hash DAG, used to represent the verified fragments that can serve as the basis for F_k in terms of time sequence and business state transition; e is the business event identifier parsed from the logistics business content carried by the fragment; H(C_j) is the digest value obtained by performing a secure hash operation on the complete frame content C_j of the parent fragment F_j, where C_j includes a fixed header, an extended header, an encrypted payload, and an authentication tag.

[0029] The verification rule for the DAG is: when the acquired current state of the maintenance state machine of the server is met, upon receiving , it is required that all parent shards of have passed verification and for a certain holds, otherwise the shard is discarded.

[0030] It should be noted that before constructing the causal hash DAG, a service state machine corresponding to the logistics intelligent entity session is first defined as: M=(Q,Σ,δ); Wherein, Q={demand identification q0, intelligent order placement q1, carrier order acceptance / collection q2, in-transit positioning q3, arrival at port / warehousing q4, outbound / discharging q5, signature acceptance / check and acceptance q6, reconciliation / settlement q7}; Σ is a set of logistics service events, including events of quotation confirmation, order placement, vehicle order acceptance, cargo loading and collection, trajectory reporting, warehousing, outbound, discharging, signature acceptance, receipt, cost confirmation, reconciliation, and payment settlement; δ:Q×Σ→Q is a state transition function, which is determined by the workflow rules of the platform OMS / TMS / WMS and the supply chain control tower, and is used to define the reachable transition relationship between different service phases.

[0031] In comparison with the D4 linear hash chain: even if an attacker obtains the complete shard sequence of the "demand identification / quotation phase", the attacker cannot replay it to the "reconciliation / settlement phase", because δ(demand identification, settlement event) is unreachable in the state machine, and the BC-DAG dependency verification will inevitably fail.

[0032] Schematic illustration: q0(demand identification)→q1(intelligent order placement)→q2(carrier order acceptance / collection)→q3(in-transit positioning)→q4(arrival at port / warehousing)→q5(outbound / discharging)→q6(signature acceptance / check and acceptance)→q7(reconciliation / settlement); the "abnormity reporting" event is allowed to point across phases from q3 to q7, but ordinary service segments are not allowed to have cross-phase parent-child relationships.

[0033] When the client is ready to send the k-th shard, the service content carried by this shard corresponds to a certain state. For example, for a shard that sends "the vehicle has departed, and the current positioning is at longitude X° and latitude Y°", its state identifier is q3(in-transit positioning). The client traverses all sent shards where j<k, queries the state q(F_j) of each shard F_j, and if there exists an event e such that δ(q(F_j),e)=q3, F_j is added to Parents(F_k). For the aforementioned q3 shard, its parent nodes include the shard with the state of carrier order acceptance / collection q2 and the content of "driver has accepted the order", and also include the previous shard with the state of q3(in-transit positioning).

[0034] For each parent node F_j, the client calculates its hash value H(C_j), where C_j is the complete content of F_j, including the fixed header, extended header, cryptographic payload, and authentication tag; the shard index j and the hash value are combined into a tuple (j, H(C_j)), and all tuples form a BC-DAG Pointer Set, which is written into the extended header of F_k.

[0035] The server received Then, maintain a hash table to record the indexes of verified shards and their hash values. (Analysis) Then, for each Check: Does index j exist in the table and has a matching hash value? If not, discard the table. Then, the validity of the state is checked: the server maintains the "expected state set" for the current session (which may have multiple parallel branches). If a state of a certain parent shard exists... Make = Established The event identifiers derived from parsing the fragmented content are then validated, and the server's current state is updated accordingly. Otherwise discard.

[0036] In one embodiment of this application, the zero-knowledge scenario attribute proof ZK-SAP is used to prove that the business attributes of the current session meet a preset logistics access policy without disclosing the plaintext attributes of goods, vehicles, drivers, routes, and licenses. The constraint predicate Φ of the ZK-SAP is defined as follows: ; Among them, CargoValue is the value of the goods, HazardLevel is the hazard level, RouteRisk is the route risk, PartyCount is the number of participants, VehicleType is the vehicle type, License is the vehicle or driver's license information, Route is the transportation route information, RestrictedZone is the restricted zone information, and DriverLicenseExpiry is the driver's license expiry date. This is an empty set. ValidLicenses is a set of valid licenses maintained and verified by the platform.

[0037] It should be noted that after obtaining the constraint predicate Φ, the client uses the Groth16 algorithm to generate a zk-SNARK proof π, proving that it possesses a witness x that satisfies Φ. The proof has a fixed size and is placed in the SSE extension header. The client then writes this π into the ZK-SAP Proof field of the SSE frame extension header. Simultaneously, to resist replay attacks, the π is bound to the current session's nonce and timestamp as common input.

[0038] The server receives the SSE frame, extracts π from the extended header, verifies π using the public verification key vk, and does not read any specific values ​​of attributes in the witness x throughout the process.

[0039] In existing technologies, D6 verifiers must read the plaintext of attributes before verifying the signature, which exposes sensitive attributes to the server. This application uses zk-SNARK to completely avoid reading attributes and only verifies the assertion that "the carrier has a set of attributes that satisfy the admission policy".

[0040] In one embodiment of this application, the objective function of the Rényi entropy is: ; Where L is a random variable representing the slice length. It is second order entropy; The formula for calculating the fill length is as follows: ; in, It is the fill length. It is the plaintext length. It is an empirical distribution. It is a set of length buckets, and B = {256, 512, 1024, 2048, 4096}; The padding bytes of the Rényi entropy are generated from a DK-derived PRF, with a padding length l_pad=L. m, and write l_pad to the EAFL Control field of the extension header. After decryption, extract the valid bytes according to the plaintext length m.

[0041] It should be noted that the client initializes the frequency array freq, which contains 5 counting elements freq[0] to freq[4], corresponding to length buckets B[0]=256 bytes, B[1]=512 bytes, B[2]=1024 bytes, B[3]=2048 bytes, and B[4]=4096 bytes respectively. The count of each bucket is initially 0.

[0042] In one embodiment of this application, step S4 includes the following steps: S4.1: Compile the preset business access policy into an arithmetic circuit to generate the corresponding proof key and verification key; S4.2: The client obtains its own business attribute witness and uses the proof key and Groth16 zero-knowledge proof algorithm to generate a fixed-size proof π; S4.3: Write the proof π into the extended header of the SSE frame.

[0043] It should be noted that the server uses the verification key to verify the validity of the proof π without decrypting the business attribute witness, in order to determine whether the current session meets the business access policy.

[0044] This application uses the Groth16 algorithm to generate a proof key pk and a verification key vk, wherein the proof key pk is distributed to all clients and the verification key vk is deployed on the server.

[0045] When a client initiates a session that requires compliance verification, it collects its own attribute witness x: x.HazardLevel = 4 (Class 4 Flammable Solids); x.VehicleType = "T3" (Special tanker truck for hazardous chemicals); x.License = "A12345" (This license plate is in the Merkle tree of valid license plates issued by the Ministry of Transportation). x.Route = A sequence of GPS route points that, as proven by map matching, does not pass through any marked RestrictedZone; x.DriverLicenseExpiry = "2027-12-31".

[0046] The client runs the Groth16 proof generation algorithm Prove(pk, x), outputting the proof π, a fixed value of 288 bytes. This proof π does not contain any plaintext fields of x; it is merely cryptographic evidence that an arithmetic relation holds.

[0047] In one embodiment of this application, the SSE frame includes a fixed header, an extended header, and an encrypted payload; The fixed header carries the encrypted packets of the n key shares, and the extended header carries the pointer set of the business causal hash DAG, the zero-knowledge scenario attribute proof, and the Rényi entropy filling control information. The SSE frame structure is as follows: .

[0048] It should be noted that,

[0049] (DK, PaddedPlaintext, AAD=Extension Header).

[0050] GCM comes with a built-in 128-bit authentication tag. AAD overlays the extension header to prevent tampering. The SSE frame fixed header is designed to be 2048 bits to ensure that the server can parse it quickly. The extension header centrally stores the threshold share aggregation result, zero-knowledge proof, causal topology pointer and entropy control bit, which not only retains the low latency of streaming, but also achieves security semantics that are completely different from existing technologies.

[0051] like Figure 2 As shown, in one embodiment of this application, an agent SSE communication system based on multi-party threshold derivation and business causal DAG is also provided, which, when applied to any of the foregoing methods, includes: Client 100 and server 200 are communicatively connected, and client 100 includes: The Multi-Party Threshold Dynamic Key Derivation (MTKD) module is used to dynamically determine the threshold threshold t for Shamir threshold secret sharing based on the scenario sensitivity parameter, and to split the session key DK into n key shares and distribute them to different partner entities. Business Causal Hash DAG Module; used to establish hash dependency topology between SSE shards based on a preset business state machine; The ZK-SAP module for zero-knowledge scenario attribute proofs is used to generate zero-knowledge proofs that demonstrate that business attributes meet admission policies. The Rényi entropy adaptive piecewise filling (EAFL) module is used to dynamically determine the filling length based on the second-order Rényi entropy target. A frame assembly and encryption module; used to construct an improved SSE frame containing a fixed header, an extended header, and an encrypted payload, and to configure the fixed header to carry the n key shares.

[0052] It should be noted that the input content is encrypted through 100 different modules on the client side, thereby ensuring the confidentiality of the communication.

[0053] The Multi-Party Threshold Dynamic Key Derivation (MTKD) module comprises a share management submodule and a key reconstruction submodule. The share management submodule maintains its own locally stored share (typically s1) and network addresses pointing to the share services of other collaborating parties. The key reconstruction submodule implements the Lagrange interpolation algorithm and is responsible for collecting online shares and reconstructing the session key DK. For each application layer message to be sent, the business causal hash DAG module maps it to a state of the state machine, calculates its parent node hash set, and outputs the DAG pointer set. The ZK-SAP module for zero-knowledge scenario attribute proof has built-in proof circuits and proof keys for different compliance scenarios, thereby selecting the corresponding proof key according to the current scenario and collecting locally stored user attribute witnesses; The Rényi Entropy Adaptive Fragmented Filling (EAFL) module maintains a length bucket frequency table and provides an interface for generating fill bytes. The frame assembly and encryption module is responsible for assembling the fixed header, extended header, and encrypted payload into a binary frame bit by bit.

[0054] In one embodiment of this application, the server 200 includes: A frame parsing and decryption module is used to parse the fixed header, reconstruct the session key DK, and decrypt the encrypted payload. Causality verification module; used to verify the service continuity of the fragment sequence based on the DAG pointer set and the service state machine; Zero-knowledge verification module; used to verify the zero-knowledge proof to confirm the compliance of business attributes.

[0055] It should be noted that after the client 100 executes step S8, the server 200 receives the SSE frame, parses the fixed header of the SSE frame to recover the session key DK, uses the session key DK to decrypt the encrypted payload, and performs anti-replay verification and business compliance verification based on the information in the extended header.

[0056] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A smart agent SSE communication method based on multi-party threshold derivation and business causal DAG, characterized in that, include: S1: Obtain the scenario sensitivity parameters of the session, and dynamically determine the threshold t of the Shamir threshold secret sharing scheme based on the scenario sensitivity parameters; S2: Generate a session key DK. Based on the threshold t and the multi-party threshold key derivation MTKD mechanism, the session key DK is split into n key shares, and the n key shares are distributed to different cooperating entities. S3: Construct a business causal hash DAG; S4: Generate zero-knowledge scenario attribute proofs ZK-SAP; S5: Based on the Rényi entropy adaptive fragmentation and padding EAFL mechanism, the padding length is dynamically determined and the padding operation is performed according to the length of the current plaintext data to be sent and the preset padding length probability distribution target. S6: Using the session key DK, perform symmetric encryption on the padded plaintext data to generate an encrypted payload; S7: Assemble improved SSE frames; S8: Send the SSE frame via the SSE protocol.

2. The agent SSE communication method based on multi-party threshold derivation and business causal DAG as described in claim 1, characterized in that: The threshold t is positively correlated with the scene sensitivity parameter, and the formula for calculating the threshold t is: ; Where n is the total number of cooperating entities participating in the key share distribution, σ(·) is the Sigmoid function used to map the comprehensive risk score to the interval between 0 and 1, w1~w4 are weight coefficients preset or trained based on historical waybills, vehicle trajectories, warehousing and settlement data; CargoValue is the normalized index of the value of goods corresponding to the current logistics session, HazardLevel is the cargo hazard level index, RouteRisk is the transportation route risk index, and PartyCount is the number of entities participating in this business collaboration.

3. The agent SSE communication method based on multi-party threshold derivation and business causal DAG according to claim 2, characterized in that: The derivation formula for the key DK in step S2 is: ； Where, λ i (0) represents the Lagrange interpolation coefficients, R represents the set of real-time online partners that have passed identity verification, t represents the threshold, and DK represents the session key; HKDF represents the key derivation function based on hash message authentication code, SHA3 represents the SHA3-512 secure hash algorithm, SortAttr(A) represents the information string obtained by sorting and normalizing the attribute list A according to the preset field encoding, and Nonce‖TimeSlot represents the salt value formed by concatenating the random number Nonce with the current time slice TimeSlot; any share leak less than t cannot restore DK, and the attribute list A only participates in the derivation as the Info parameter of HKDF but is not transmitted in plaintext.

4. The agent SSE communication method based on multi-party threshold derivation and business causal DAG as described in claim 3, characterized in that: The nodes of the DAG represent fragments of the SSE protocol, and the directed edges of the DAG represent the legal state transition relationships between adjacent fragments based on a preset service state machine. The construction rule of the DAG is as follows: for the k-th partition The state of being The set of parent nodes is: In the fragmentation extension header ; Wherein, Parents(F_k) is the set of parent nodes of the k-th SSE fragment F_k in the business causal hash DAG, used to represent the verified fragments that can serve as the basis for F_k in terms of time sequence and business state transition; e is the business event identifier parsed from the logistics business content carried by the fragment; H(C_j) is the digest value obtained by performing a secure hash operation on the complete frame content C_j of the parent fragment F_j, where C_j includes a fixed header, an extended header, an encrypted payload, and an authentication tag. The verification rule for the DAG is as follows: when the current state of the server's maintenance state machine is obtained... At that time, received At that time, it was required All parent fragments have been verified and For a certain If valid, otherwise discard.

5. The agent SSE communication method based on multi-party threshold derivation and business causal DAG according to claim 4, characterized in that: The ZK-SAP proof is used to prove that the service attributes of the current session meet a preset admission policy without disclosing the attribute plaintext. The constraint predicate of the ZK-SAP proof... Defined as: ； Among them, CargoValue is the value of the goods, HazardLevel is the hazard level, RouteRisk is the route risk, PartyCount is the number of participants, VehicleType is the vehicle type, License is the vehicle or driver's license information, Route is the transportation route information, RestrictedZone is the restricted zone information, and DriverLicenseExpiry is the driver's license expiry date. An empty set indicates that the transport route does not intersect with the restricted area; ValidLicenses is a set of valid licenses and permits on the platform, including vehicle road transport permits, driver qualification certificates, carrier operating licenses, and hazardous or special transport qualifications, which are written into the Merkle tree after being reviewed by the platform, filed with the road administration, or verified by a third-party interface.

6. The agent SSE communication method based on multi-party threshold derivation and business causal DAG according to claim 5, characterized in that: The objective function for the Rényi entropy is: ; Where L is a random variable representing the slice length, It is second order entropy; The formula for calculating the fill length is as follows: ； in, It is the fill length. It is the plaintext length. It is an empirical distribution. It is a set of length buckets, and B = {256, 512, 1024, 2048, 4096}; The padding bytes of the Rényi entropy are generated from a DK-derived PRF and are set with length... Write to the EAFLControl field in the extended header, and after decryption, directly extract the first m bytes.

7. The agent SSE communication method based on multi-party threshold derivation and business causal DAG according to claim 6, characterized in that: Step S4 includes the following steps: S4.1: Compile the preset business access policy into an arithmetic circuit to generate the corresponding proof key and verification key; S4.2: The client obtains its own business attribute witness and uses the proof key and Groth16 zero-knowledge proof algorithm to generate a fixed-size proof π; S4.3: Write the proof π into the extended header of the SSE frame.

8. The agent SSE communication method based on multi-party threshold derivation and business causal DAG according to claim 7, characterized in that: The SSE frame includes a fixed header, an extended header, and an encrypted payload; The fixed header carries the encrypted packets of the n key shares, and the extended header carries the pointer set of the business causal hash DAG, the zero-knowledge scenario attribute proof, and the Rényi entropy filling control information. The SSE frame structure is as follows: 。 9. An agent SSE communication system based on multi-threshold derivation and business causal DAG, applied to the method of any one of claims 1 to 8, characterized in that, include: The client and server are connected for communication. The client includes: The Multi-Party Threshold Dynamic Key Derivation (MTKD) module is used to dynamically determine the threshold threshold t for Shamir threshold secret sharing based on the scenario sensitivity parameter, and to split the session key DK into n key shares and distribute them to different partner entities. Business Causal Hash DAG Module; used to establish hash dependency topology between SSE shards based on a preset business state machine; The ZK-SAP module for zero-knowledge scenario attribute proofs is used to generate zero-knowledge proofs that demonstrate that business attributes meet admission policies. The Rényi entropy adaptive piecewise filling (EAFL) module is used to dynamically determine the filling length based on the second-order Rényi entropy target. A frame assembly and encryption module; used to construct an improved SSE frame containing a fixed header, an extended header, and an encrypted payload, and to configure the fixed header to carry the n key shares.

10. The agent SSE communication system based on multi-threshold derivation and business causality DAG according to claim 9, characterized in that, The server includes: A frame parsing and decryption module is used to parse the fixed header, reconstruct the session key DK, and decrypt the encrypted payload. Causality verification module; used to verify the service continuity of the fragment sequence based on the DAG pointer set and the service state machine; Zero-knowledge verification module; used to verify the zero-knowledge proof to confirm the compliance of business attributes.

Images